Michael Christopher

This mole looks completely normal. But, it’s not a mole. It invasive melanoma. Yes, this little spot. And when you get a skin check, if they don’t put a dermatoscope on every mole, every sunspot, and every pink spot, a melanoma like this can be easy to miss. Standards have

Why do I keep posting cases about dermatologists missing melanoma, and not using dermoscopy? Watch this video. This is not isolated. Some dermatologist don’t use dermoscopy at all. Some only use it on spots that concern them. It needs to be applied to every mole, every

@exosome Oh sure. Like run the experiment- in real life. Measure the outcome and then see what the results show? lol. This hyperbole is wild lol

Today we all lost our jobs..... Three Nature papers showing that scientists in the conventional sense are obsolete At least read the first one.... the AI replaced all things that the scientist does .... nature.com/articles/s4158…

@raphaels7 @theproof You claimed atherosclerosis can occur without plaque. I specifically stated to you atherosclerosis has to have plaque, by definition. And you are now going to your “like most doctors” BS, as if you harbor knowledge- which you do not, but, decide to claim for validation.

@MChristopher_MD @theproof Yes, not visible on angiography, which is how approximately 85-90% of plaque is detected in the real world, but potentially visible via IVUS and certainly via autopsy. Like most doctors, you fail to grasp that "All models are wrong, but some are useful"

Funny that: he’ll “dig into” validation when results show keto is awesome, but takes everything at face value when a shitty AI reports keto is harmful

@raphaels7 @theproof Did you read it? You still do not understand the fundamental basics of an argument you’re trying to make. Ischemia can occur in absence of atherosclerosis. But, atherosclerosis fundamentally has to have plaque- to be defined as atherosclerosis.

@MChristopher_MD @theproof pmc.ncbi.nlm.nih.gov/articles/PMC60… you’re welcome

@theproof @raphaels7 He made a claim that you can have an atherosclerotic event without plaque. I do not believe he understands the fundamental basic understanding of what atherosclerosis is. I am all for discourse- but, some points invalidate the rest, like this.

@raphaels7 To cut to the chase. I don't think they are. So we are assuming they are reliable and we aren't just seeing results affected by noise. And we have no idea what timeframe is needed to reliably detect change in these people.

@raphaels7 @theproof What in the world is this nonsense answer? You absolutely cannot have atherosclerosis without plaque. You can have an ischemic event without plaque, and that is not the same thing. Seriously, you can not take yourself seriously with that nonsense. @ethanjweiss

@theproof OK, so plaque isn’t strictly necessary for atherosclerosis, as (i think we agree) you can have an atherosclerotic event without it. So back to plaque: any type of plaque >0?

Melanomas are missed far too often during routine skin exams. How do I know? Here is an example of a large, almost 4 cm patch that was being frozen by a board certified dermatologist, because they thought it was a sun spot. And that is not acceptable. We need to change the

Melanoma can look normal. This melanoma looks like a normal mole. And even though this spot is pink, it’s not an outlier on this patient. When you look inside of it, that is the only way you would know this spot is concerning and melanoma. With dermoscopy education and

Why did I make a case series of melanoma not being obvious for melanoma awareness month? Look at the level of sun damage present on this neck? There are numerous pink spots, black spots, and irregular brown spots. There is only one way to tell what is normal, and what is

@ethanjweiss Through tens of thousands of reps- and even more reps on the back end, that no one will ever see.

Experience matters. Training matters. Years of failure and iteration matter. Expertise is earned, not performed

Melanoma can occur inside the eye, not just on your skin. Most people think of melanoma as only skin cancer, but melanocytes are present in the eye, specifically in a layer of the eye called the uvea. The uvea includes: • the iris- the colored part you can see • the ciliary

@ethanjweiss @theproof Spot on Ethan.

CAC/plaque is a noisy biomarker at the low end of the spectrum — this study was always going to be hard to pull off. Here's how I would have designed it: enroll only people who demonstrate a meaningful LDL-C/ApoB rise in response to a ketogenic diet AND have visible atherosclerosis on CTA at baseline. No algorithm required — just real phenotypic enrichment. Phase 1 (1 month): deliver all meals to minimize heterogeneity, then document that LDL-C/ApoB actually rises meaningfully on keto. Confirm the phenotype is real before you randomize anyone. Phase 2 (1 year): take the confirmed responders and randomize 1:1 — keto diet vs. a matched diet with carb and fat macros swapped. CTA at one year. Primary endpoint: difference in plaque volume between groups. Secondary: within-group change from baseline. The signal you're looking for — if it exists — lives in people who are both diet-responsive and already have plaque. Enrich for both or you're underpowered before the first patient is enrolled. Again this is a hard study but if you want to make anything of the results, it is a good idea to start with something that can be interpreted. The existing design was always shit and I have been transparent about this consistently from the beginning. Wake me when there is a real experiment to discuss

Worth noting for everyone interested in the Keto-CTA and LMHR research (I include myself in this group and am staying curious). Are QAngio, Cleerly or Heartflow validated in asymptomatic subjects with sub-clinical atherosclerosis? Short answer: no — not in a true subclinical-atherosclerosis cohort. All three platforms have published head-to-head IVUS comparisons, but every one of those studies enrolled patients who were clinically referred for invasive coronary imaging (i.e. known or suspected CAD), because you can't ethically push an IVUS catheter into an asymptomatic person just to validate a CT algorithm. This means they have not been compared to gold standard for people with minimal plaque where there is likely much more measurement noise. Each of the three has IVUS-validated its plaque quantification, but only in symptomatic / known-CAD patients undergoing invasive imaging — none has a published head-to-head IVUS validation in a true subclinical-atherosclerosis cohort, and that gap is fundamentally ethical (you don't IVUS asymptomatic people) rather than something likely to be filled. Why the gap matters: in subclinical disease the plaques are typically smaller, more often non-calcified, and located in less-stenotic segments — exactly the regime where CCTA's spatial resolution is most stretched. Algorithms tuned and validated on bulky, stenotic, IVUS-imaged plaques may not perform identically on a CAC-zero, mildly diseased asymptomatic vessel. To be clear this isn't an argument for or against the KETO-CTA study results. It's a question - are we using the right tool to determine risk? The answer is ... a tool is being used that's not validated in asymptomatic people with sub-clinical atherosclerosis. Therefore we need to be really careful what we make of the results and how it's used to patient recommendations. I'll be digging much deeper into the Pre-print from this group and AI analysis of plaque in asymptomatic subclinical atherosclerosis in the coming weeks with several esteemed guests.

Would you think this spot on the back of the ear is melanoma? This spot is light, does not stand, and looks like sun damaged skin. But this little spot, this is melanoma. This is why your skin exam requires dermoscopy (the zoomed in view) on every pink and pigmented spot.

Normal appearing mole, or is it actually normal? This is not normal. This is melanoma. Want to understand why? Follow me. I am shining light on a massive flaw in our current standard for skin cancer screening. #melanoma #melanomaawareness

Melanoma is not obvious, Case 6 in a series of cases for melanoma awareness month. Why this matters- melanoma is the most deadly form of skin cancer, and understanding the shift needed for early detection starts with seeing why most melanomas are- not obvious. This spot is

Melanoma is not obvious, Case 5 in a series of cases for melanoma awareness month. Why this matters- melanoma is the most deadly form of skin cancer, and understanding the shift needed for early detection starts with seeing why most melanomas are not obvious. This spot is

Dr. Adamson highlighted exactly the point I have been trying to make for years. There is a massive detection gap in melanoma. Let’s provide some context. The melanoma numbers provided are from the Arizona department of health. My patient population is the same as everyone

Melanoma awareness month, case 4. This spot is symmetric. Has regular borders. And is one color. It is melanoma. And this melanoma would be overlooked during most skin exams in the United States. This is your reminder, skin exams require dermoscopy on every mole, not just

@FCademartiri This whole post is AI. Wrote by AI. Honestly- it’s slop.

🤖🏥 Everyone is talking about AI in healthcare. Almost nobody is talking about what happens when AI starts acting autonomously. And that’s the real revolution. Not chatbots. Not prompts. Not “write my discharge summary.” 👉 Agents. AI systems that: - reason - plan - use tools - collaborate - self-correct - execute multi-step workflows autonomously This review makes one thing very clear: We are moving from AI as software… to AI as digital workforce. Let’s be honest Most healthcare systems are drowning in: - fragmentation - overload - bureaucracy - cognitive fatigue - disconnected data silos And humans are currently acting as: biological middleware. AI agents change that completely Not because they “know medicine.” But because they can orchestrate: 👉 imaging 👉 EHRs 👉 guidelines 👉 reports 👉 workflows 👉 triage 👉 communication 👉 decision pathways —in real time. The scary part? Some systems are already performing at near-specialist level. Examples in the paper include: - multi-agent oncology MDTs - cardiology diagnostic systems - autonomous report generation - AI-driven emergency triage - pharmacovigilance orchestration - AI patient simulations for medical education Translation Medicine is slowly becoming: 👉 an agentic ecosystem Where AI no longer “assists” isolated tasks… but coordinates entire clinical processes. But here’s the uncomfortable truth Healthcare is NOT a benchmark dataset. It is: ❌ ambiguity ❌ incomplete information ❌ liability ❌ emotion ❌ biological variability ❌ human trust And the paper repeatedly highlights the real dangers: ⚠ hallucinations ⚠ accountability gaps ⚠ bias ⚠ lack of interpretability ⚠ over-reliance My take The biggest misconception is this: 👉 people think AI agents are replacing physicians. Wrong. They are replacing: fragmentation. The real paradigm shift From: ❌ isolated software tools To: 👉 autonomous clinical orchestration systems And radiology? Radiology is probably ground zero. Because imaging already sits at the intersection of: - data - workflow - AI - multimodality - triage - prediction - decision support Which means radiologists may evolve from: 👉 image readers to: 👉 managers of diagnostic intelligence ecosystems Final thought The future hospital may not run on: 👉 departments It may run on: interconnected AI agents supervised by humans. ⚡ The future of healthcare is not “AI-powered.” It is: AI-orchestrated. And most people still underestimate how big that shift really is. #AI #Healthcare #AIAgents #DigitalHealth #Radiology #MedicalAI #FutureOfMedicine #PrecisionMedicine #LLM